US20160335693A1 - Information providing device, information providing method, program and non-transitory recording medium - Google Patents

Abstract

An obtaining unit (120) obtains a ranking of a commercial product belonging to a category defined on an e-marketplace. A collecting unit (121) collects a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed. A calculating unit (122) calculates a score of each collected term based on at least a number of collections of the term. A presenting unit (123) presents, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score in an upper ranking position.

Description

TECHNICAL FIELD
• The present disclosure relates to an information providing device, an information providing method, a program and a non-transitory recording medium.
BACKGROUND ART
• Conventionally, commercial product search devices are known which present, to a customer, information on commercial product, in order to further refine a search result on a shopping site or an auction site over the Internet, and which assist the search refinement (see, for example, Patent Literature 1).
CITATION LISTPatent Literature
• Patent Literature 1: Unexamined Japanese Patent Application Kokai Publication No. 2012-194685
SUMMARY OF INVENTIONTechnical Problem
• However, information on commercial product presented by such commercial product search devices is not weighted in accordance with the content shown by such information, and pieces of searched information on the commercial products are likely to be presented equally. Hence, information on the commercial products to be presented often contains improper terms for search refinement, such as a generic name and a category name.
• The present disclosure has been made in view of the aforementioned circumstances, and it is an objective of the present disclosure to provide an information providing device, an information providing method, a program, and a non-transitory recording medium which can present, to a customer, a suggested keyword appropriate as a term for search refinement.
Solution to Problem
• In order to accomplish the above objective, an information providing device according to a first aspect of the present disclosure includes:
• • an obtaining unit that obtains a ranking of a commercial product belonging to a category defined on an e-marketplace;
  • a collecting unit that collects a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;
  • a calculating unit that calculates a score of each collected term based on at least a number of collections of the term; and
  • a presenting unit that presents, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position.
• In the above information providing device:
• • in the e-marketplace, a category hierarchy may be defined by a tree structure with each category disposed at a node; and
  • the score of the collected term for the category may be defined based on a product:
  • (a) a weighting defined by the number of collections of the term; and
  • (b) a contrary category frequency defined by a number of child categories of the category, and a number of child categories where the term is collected among the child categories.
• In the above information providing device, the score of the collected term for the category may be defined based on a value obtained by decreasing the number of collections of the term for the category on a basis of a number of collections of the term in another category.
• In the above information providing device, a higher a similarity between the category and the other category is, a smaller a degree of decrease may become.
• In the above information providing device:
• • in the e-marketplace, a category hierarchy may be defined by a tree structure with each category disposed at a node; and
  • in the tree structure, when the category and the other category are brotherhood categories, a similarity between the category and the other category may be high.
• In the above information providing device:
• • in the e-marketplace, a category hierarchy may be defined by a tree structure with each category disposed at a node; and
  • in the tree structure, when a distance between the category to the other category is short, a similarity between the category and the other category may be high.
• In the above information providing device:
• • in the e-marketplace, each of a plurality of shops may create the sales page, and the same commercial product belonging to the same category may be available from the respective sales pages; and
  • the ranking may be obtained for each category to which the sold commercial product belongs regardless of from which shop the commercial product is sold.
• In order to accomplish the above objective, an information providing method according to a second aspect of the present disclosure is executed by an information providing device including an obtaining unit, a collecting unit, a calculating unit, and a presenting unit, and, the method includes:
• • obtaining by the obtaining unit a ranking of a commercial product belonging to a category defined on an e-marketplace;
  • collecting by the collecting unit a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;
  • calculating by the calculating unit a score of each collected term based on at least a number of collections of the term; and
  • presenting by the presenting unit, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position.
• In order to accomplish the above objective, a program according to a third aspect of the present disclosure causes a computer to function as:
• • an obtaining unit that obtains a ranking of a commercial product belonging to a category defined on an e-marketplace;
  • a collecting unit that collects a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;
  • a calculating unit that calculates a score of each collected term based on at least a number of collections of the term; and
  • a presenting unit that presents, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position.
• In order to accomplish the above objective, a non-transitory computer-readable recording medium according to a fourth aspect of the present disclosure has stored therein a program that causes a computer to function as:
• • an obtaining unit that obtains a ranking of a commercial product belonging to a category defined on an e-marketplace;
  • a collecting unit that collects a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;
  • a calculating unit that calculates a score of each collected term based on at least a number of collections of the term; and
  • a presenting unit that presents, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position.
• The above program may be distributed and sold separately from a computer that executes the program over a computer communication network. In addition, the above recording medium may be a non-transitory recording medium that can be separately distributed and sold from the computer.
Advantageous Effects of Invention
• According to the information providing device, the information providing method, the program and the non-transitory recording medium of the present disclosure, it becomes possible to provide, to a customer, a suggested keyword appropriate as a term for search refinement.
BRIEF DESCRIPTION OF DRAWINGS
• FIG. 1 is a diagram illustrating a structure of an information providing system according to an embodiment of the present disclosure;
• FIG. 2 is a schematic block diagram illustrating a hardware structure of the information providing device;
• FIG. 3 is a schematic block diagram illustrating a functional structure of the information providing device;
• FIG. 4 is a diagram illustrating an example category tree structure;
• FIG. 5 is a diagram illustrating example data stored in a commercial product database;
• FIG. 6 is a diagram illustrating an example purchase table;
• FIG. 7 is a diagram illustrating an example category table;
• FIG. 8 is a diagram illustrating an example ranking table;
• FIG. 9 is a diagram illustrating an example sales page;
• FIG. 10 is a diagram illustrating an example table of a relevant term for a given category;
• FIG. 11 is a diagram illustrating an example table of a relevant term for another category;
• FIG. 12 is a diagram illustrating an example table of appearance frequency of a relevant term;
• FIG. 13 is a diagram illustrating another example table of appearance frequency of a relevant term;
• FIG. 14 is a diagram illustrating an example ranking page;
• FIG. 15 is an example flowchart of an information providing process;
• FIG. 16 is a diagram illustrating an example tree structure for women's fashion;
• FIG. 17 is a diagram illustrating another example table of appearance frequency of a relevant term;
• FIG. 18 is a diagram illustrating an example table showing a contrary category frequency;
• FIG. 19 is a diagram illustrating an example table showing a calculated weighting value based on a contrary category frequency;
• FIG. 20A is a diagram illustrating an example score transition of a relevant term;
• FIG. 20B is a diagram illustrating an example score transition of a relevant term; and
• FIG. 20C is a diagram illustrating an example score transition of a relevant term.
DESCRIPTION OF EMBODIMENTS
• Embodiments of the present disclosure will be explained below with reference to the accompanying figures.
First Embodiment
• FIG. 1 illustrates a general structure of aninformation providing system1 according to this embodiment. Theinformation providing system1 is a system that provides a suggested keyword appropriate when a customer searches a commercial product in an e-marketplace. As illustrated inFIG. 1, theinformation providing system1 includes aninformation providing device100,shop terminals200,customer terminals300, and acommercial product database400, and, the respective devices are connected so as to be communicable one another via theInternet500.
• Theinformation providing device100 is a computer system which searches commercial products in thecommercial product database400 based on a query specified by the customer, creates a ranking based on the commercial products purchased by customers from the searched commercial products, collects a term relevant to the commercial product in the ranking, and calculates a score from the collected term and presents a suggested keyword. More specifically, theinformation providing device100 obtains a ranking for the commercial product to which purchase offers are made from themultiple customer terminals300, collects a relevant term to the commercial product at the upper ranking position, calculates a score of the collected relevant term to the commercial product, and presents, to thecustomer terminal300, the upper ranking score term as the suggested keyword. Theinformation providing device100 can be realized by various devices like a server device.
• Theshop terminal200 is a computer operated by a shop that sells commercial products on the e-marketplace. Theshop terminal200 accepts an inputting of information on commercial products that the shop wants to sell, and transmits the information on the commercial products to theinformation providing device100. In addition, theshop terminal200 enables, in accordance with a request from thecustomer terminal300, the customer to view, via theInternet500, a commercial product sales page created by the shop.
• Thecustomer terminal300 is a computer operated by the customer who purchases the commercial product on the e-marketplace. Thecustomer terminal300 accepts an inputting of information on the commercial product that the customer wants to purchase, and transmits purchase offer information to theshop terminal200 through theinformation providing device100. In addition, thecustomer terminal300 displays, on a display, the suggested keyword presented by theinformation providing device100.
• Thecommercial product database400 is a device that stores information on the commercial products that the shops want to sell on the e-marketplace.FIG. 5 illustrates example data stored in thecommercial product database400. Thecommercial product database400 stores a commercial product ID, a commercial product name, a category ID, a shop ID, a commercial product price, and a commercial product description.
• The commercial product name is a name of commercial product input by the shop, and is also a name of commercial product determined by the shop arbitrarily. The commercial product ID is an identifier defined corresponding to the commercial product name, and the category ID is an identifier to identify the category to which the commercial product belongs. The shop ID is an identifier to identify the shop that sells the commercial product. The commercial product description is information relating to the commercial product and written by the shop in the sales page.
• Next, the structure of theinformation providing device100 will be explained in more detail.
• FIG. 2 is a schematic block diagram illustrating a hardware structure of theinformation providing device100. As illustrated inFIG. 2, theinformation providing device100 includes acontroller101, a Read Only Memory (ROM)102, a Random
• Access Memory (RAM)103, adisplay104, acommunicator105, and anoperation hardware106, and the respective elements are connected one another via abus107.
• Thecontroller101 includes, for example, a Central Processing Unit (CPU), and controls the wholeinformation providing device100.
• TheROM102 is a non-volatile memory that stores a program and various data for thecontroller101 to control the wholeinformation providing device100.
• TheRAM103 is a volatile memory that temporarily stores information created by thecontroller101, and data necessary to create such information.
• Thedisplay104 is a display device that includes an LCD (Liquid Crystal Display), a backlight, and the like. Thedisplay104 displays, for example, data output by thecontroller101 under the control thereof.
• Thecommunicator105 includes a communication interface to connect theinformation providing device100 to theInternet500.
• Theoperation hardware106 includes input devices, such as buttons, a touch panel, and a keyboard. Theoperation hardware106 receives an input operation from the customer or the shop, and outputs, to thecontroller101, an input operation signal corresponding to the received input operation.
• Next, a functional structure of theinformation providing device100 will be explained.
• FIG. 3 is a schematic block diagram illustrating a functional structure of theinformation providing device100. As illustrated inFIG. 3, theinformation providing device100 includes an obtainingunit120, a collectingunit121, a calculatingunit122, and a presentingunit123.
• The obtainingunit120 obtains a ranking that belongs to a category defined on the e-marketplace. The obtainment of the ranking starts when theinformation providing device100 creates a purchase table based on the commercial product with a purchase procedure completed. In this case, the e-marketplace is a marketplace that enables a business-to-business transaction over the Internet. In addition, the ranking is obtained for each category defined on the e-marketplace.
• As for the category structure of the commercial products on the e-marketplace, respective categories may be arranged in a parallel relation one another, or a tree structure in which each category is disposed with a node may be employed. In this embodiment, an explanation will be given of a category structure in which the categories are hierarchically arranged as a tree structure.FIG. 4 is a diagram illustrating an example category structure that is a hierarchical tree structure. Atree structure401 of the categories illustrated inFIG. 4 is a tree structure for a category that is women's fashion. In thistree structure401, women's fashion is a parent node that is a root, and this parent node is linked with multiple child nodes that are one-piece suit, dress, and suit with edges. The multiple child nodes are in a brotherhood relationship. Each child node is further linked with grandchild nodes with edges. That is, one-piece suit that is the child node is linked with grandchild nodes that are long, short, and medium with edges, and dress that is the child node is linked with grandchild nodes that are party, formal with edge.
• An explanation will be given of a process of obtaining a ranking by the obtainingunit120 with reference to an example ranking of commercial products belonging to a category that is women's fashion.
• A ranking is created upon creation of the purchase table. Hence, a process of creating the purchase table will be explained first. When receiving a search query from the customer, thecustomer terminal300 transmits the received search query to theinformation providing device100. Thecontroller101 of theinformation providing device100 searches, from thecommercial product database400 based on the received search query, the commercial products containing a letter string that matches the search query, and transmits the searched commercial products to thecustomer terminal300. When any one of the searched commercial products is selected by the customer, thecustomer terminal300 transmits, to theinformation providing device100, information to the effect that the purchase offer to the commercial product is made. Theinformation providing device100 transmits, to theshop terminal200, the information to the effect that the purchase offer to the commercial product is made, and theshop terminal200 starts a procedure for the purchase offer. In theshop terminal200, after the purchase procedure for the customer completes, information on the purchased commercial product is transmitted to theinformation providing device100 from theshop terminal200, and is stored in the purchase table of theinformation providing device100. This process is performed for eachcustomer terminal300.
• FIG. 6 illustrates an example purchase table600 stored in theinformation providing device100. The purchase table600 stores a commercial product ID of the commercial product purchased by the customer, a commercial product category ID to which this commercial product belongs, the commercial product price, the purchase quantity purchased within a predetermined time period, and the search query applied to search the commercial product.
• The commercial product category ID is a reference index for a category classified and defined in accordance with the attribute of the commercial product, and is stored in a category table in association with the category name.FIG. 7 illustrates an example category table700 showing a correspondence relationship between the category name and the category ID in the category that is women's fashion. A category ID J001 is given to the uppermost layer category in the category structure which is women's fashion, category IDs J011, J012, and J013 are given to the intermediate layer categories that are one-piece suit, dress, and suit, and category IDs J111, J112, J113, J211, and J212 are given to the lowermost layer categories that are long, short, medium, party, and formal.
• The category ID stored in the purchase table600 and the category ID stored in thecommercial product database400 are the consistent category ID. Hence, thecommercial product database400 and the purchase table600 are tied up by the category ID, and thus, as will be discussed later, when a term relevant to a commercial product in a specific category is collected, textural information stored in both database and table can be collected as the term relevant to the commercial product.
• The ranking is determined based on a predetermined time period, and for example, any time period among real time, day by day, week by week, month by month, and year by year is selected by the customer. Accordingly, the time period of the ranking is determined. In a specific category, in accordance with the time period of the ranking selected by the customer, theinformation providing device100 creates the purchase table600, and the obtainingunit120 obtains the ranking based on the created purchase table600.
• More specifically, the obtainingunit120 calculates, for each commercial product within the predetermined time period, a total amount of sales, that is, a value obtained by multiplying the sales price by the sales quantity using data in the purchase table600. Next, the obtainingunit120 sorts and arranges the commercial products in the order of a higher total amount of sales, and creates a ranking table800 based on the amount of sales.
• FIG. 8 illustrates an example ranking table800. The ranking table800 stores the commercial product ID, a total amount of sales, and a position in the ranking. The ranking table800 is created for each predetermined category. When the category is structured hierarchically, this table may be created for each uppermost layer category, for each intermediate layer category, or for each lowermost layer category. The ranking table800 of this embodiment is a ranking table for one-piece suit that is the category J011, and the commercial products are arranged in the order of a higher total amount of sales.
• The ranking is not limited to a ranking based on the amount of sales, but may be other kinds of ranking like a ranking of popular commercial products. When a ranking of the popular commercial products is to be obtained, a ranking of the commercial products can be determined based on a table that has a reference index which is the number of popularity votes or the like.
• In this embodiment, thecontroller101, thecommunicator105, and theROM102 work together to function as the obtainingunit120.
• The collectingunit121 collects, from a text contained in a sales page for selling the upper ranking position commercial products or the search query that triggers this sales page to be viewed, a term relevant to the upper ranking position commercial products in the obtained ranking.
• An explanation will now be given below of a process of collecting a relevant term to the commercial product by the collectingunit121 with reference to a process of obtaining a relevant term to the upper-ranking commercial product in the obtained ranking in women's fashion.
• First, an explanation will be given of an example case in which a relevant term to the commercial product is collected from a text contained in the sales page for selling the commercial product.
• The sales page for selling the commercial product is a page to which information on the commercial product that the shop wants to sell is input. Normally, the sales page is created by the shop that inputs commercial product information to theshop terminal200.FIG. 9 illustrates an example sales page. Thesales page900 contains the shop name of the shop that sells the commercial product, the commercial product name, the commercial product description, and the picture of the commercial product, and the like. Among pieces of information input to thesales page900, the textural information is transmitted to theinformation providing device100 from theshop terminal200, and is stored in thecommercial product database400. More specifically, the commercial product name and the description of the commercial product that are the textural information in thesales page900 are stored in the fields of thecommercial product database400 which are the commercial product name and the commercial product description. In addition, thesales page900 may be linked with theshop terminal200 when the customer selects the commercial product that the customer wants to purchase through thecustomer terminal300, and then the customer may be enabled to view the sales page.
• The collectingunit121 refers to the textural information written in the commercial product description, the commercial product name, and the like in thecommercial product database400 for the upper position commercial product in the ranking obtained by the obtainingunit120, more specifically, for the commercial product with a ranking position that is equal to or higher than a predetermined position in the ranking table800, and collects the relevant term to the commercial product with the ranking position that is equal to or higher than the predetermined position.
• The relevant term to the commercial product may be terms divided by a minimum unit term. In the case of characters of a language like English that is divided by words, the relevant term becomes a term divided word by word. In the case of characters of a language like Japanese not having a space between words, the minimum unit term may be determined through a scheme like morphological analysis that divides characters into a string of minimum unit terms each have a meaning.
• Next, an explanation will be given of collecting a relevant term to the commercial product from the search query that triggers the sales page to be viewed.
• When the customer searches a commercial product that the customer wants to purchase, the customer inputs the search query to thecustomer terminal300, and theinformation providing device100 searches the commercial products relating to the letter string in the search query based on the input search query. When the customer purchases any commercial product among the searched commercial products, the search query corresponding to the purchased commercial product is stored in the search query field in the purchase table600 illustrated inFIG. 6. The collectingunit121 refers to the search query in the purchase table600 tied up by the commercial product ID for the commercial product with the ranking position that is equal to or higher than the predetermined position in the ranking table800, and collects the relevant term to the commercial product.
• The relevant term to the commercial product may be the search query itself or may be a divided term by the minimum unit term as explained above.
• The collected relevant term to the commercial product is stored in a relevant term table for each category.FIG. 10 andFIG. 11 show an example relevant term table collected for each category.FIG. 10 is a table of relevant term to the commercial product collected in the category that is one-piece suit, andFIG. 11 is a table of relevant term to the commercial product collected in the category that is dress.
• In this embodiment, thecontroller101, thecommunicator105, and theROM102 work together to function as the collectingunit121.
• The calculatingunit122 calculates the score of each collected term based on at least the number of collections of such a term.
• An explanation will be given of a process of calculating the score of the collected term based on the number of collections by the calculatingunit122 with reference to an example process of calculating the store based on the collected term in the category that is one-piece suit and in the category that is dress. In this case, the number of collections is, for example, the number of appearances of the relevant term to the commercial product or the number of appearances of the sales page that shows the relevant term to the commercial product.
• When the number of collections is the number of appearances of the relevant term to the commercial product, the score of such a term is obtained as an appearance frequency indicating how many times the relevant term to the commercial product appears within a predetermined time period, for example, real time, day by day, week by week, month by month, and year by year.FIG. 12 andFIG. 13 illustrate an example appearance frequency table of the relevant term.FIG. 12 is a table showing the appearance frequency of the relevant term in the category that is one-piece suit, andFIG. 13 is a table showing the appearance frequency of the relevant term in the category that is dress. The larger the number of appearances of the relevant term to the commercial product is, the larger the score of the appearance frequency becomes.
• For example, as for the appearance frequency of the relevant term in the category that is one-piece suit is, as illustrated inFIG. 12, the number of appearances within the predetermined time period is indicated as the score that is the appearance frequency for each relevant term, and the score of the appearance frequency is 2 for “gift”, 2 for “free shipping”, 3 for “mini one-piece suit”, 2 for “mini spring one-piece suit”, and 1 for “long sleeve”. In addition, the score of the appearance frequency of the relevant term in the category that is dress is, as illustrated inFIG. 13, 1 for “second party”, 2 for “free shipping”, 2 for “bridal”, 2 for “funeral ceremony”, 1 for “gift”, and 2 for “funeral”.
• When the number of collections is the number of the appearance frequency of the sales page that shows the relevant term to the commercial product, the score is determined based on how many sales pages created by the shop and showing the relevant term to the commercial product is collected within the predetermined time period. The sales page is normally created shop by shop that sells the commercial product, but the multiple sales pages may be created by the same shop.
• For example, the score is determined based on how many pages the sales page appears within the predetermined time period for the commercial product with the ranking position that is equal to or higher than the predetermined position. In thesales page900 illustrated inFIG. 9, the relevant term “mini one-piece suit” to the commercial product appears. When the term “mini one-piece suit” appears in the other sales pages, for example, asales page901, and asales page902, the number of appearances of the term “mini one-piece suit” is 3. The calculatingunit122 calculates the number of collections as 3. Note that the term “mini one-piece suit” appears twice in thesales page900, but is counted as1 for this single sales page.
• In this embodiment, thecontroller101 and thecommunicator105 work together to function as the calculatingunit122.
• The presentingunit123 presents, as a suggested keyword for commercial product search refinement belong to the category, the term with the calculated score at the upper ranking position.
• An explanation will now be given of a process of presenting the suggested keyword for search refinement by the presentingunit123 with reference to an example process of presenting the suggested keyword for the category that is one-piece suit and for the category that is dress.
• The presentingunit123 presents, as the suggested keyword for commercial product search refinement, the relevant term with the appearance frequency that is at an upper position among the relevant terms. For example, in the appearance frequency table of the relevant term illustrated inFIG. 12, as for the appearance frequency of the relevant term, the appearance frequency increases in the order of “long sleeve”→“gift”, “free shipping”, “mini spring one-piece suit”→“mini one-piece suit”. Hence, the presentingunit123 presents, to thecustomer terminal300, the relevant term with the appearance frequency that is equal to or greater than, for example, 2 as the suggested keyword for commercial product search refinement.
• The presentingunit123 presents, as the suggested keyword for commercial product search refinement, the relevant term with the appearance frequency that is at the upper position among the relevant terms. For example, in the appearance frequency table of the relevant term illustrated inFIG. 13, as for the appearance frequency of the relevant term, the appearance frequency increases in the order of “second party”, “gift”→“free shipping”, “bridal”, “funeral ceremony”, and “funeral”. Hence, the presentingunit123 presents, to thecustomer terminal300, the relevant term with the appearance frequency that is equal to or greater than, for example, 2 as the suggested keyword for commercial product search refinement.
• More specifically, the presentingunit123 provides the suggested keyword as a part of a ranking page on the display of thecustomer terminal300.FIG. 14 illustrates anexample ranking page1400 containing the suggested keyword and displayed on the display of thecustomer terminal300. Theranking page1400 contains a searchquery input part1401, a ranking timeperiod display part1402, a suggestedkeyword display part1403, and aranking display part1404. The searchquery input part1401 is a field to input the query desired by the customer for searching, the ranking timeperiod display part1402 is a part to display a target time period to create the ranking, the suggestedkeyword display part1403 is a part to display the suggested keyword for commercial product search refinement, and theranking display part1404 is a part to display the ranking of obtained based on the search query input by the customer and obtained by the obtainingunit120.
• When the customer views the ranking result displayed at theranking display part1404, and wants to perform commercial product search refinement, the customer refers to the suggested keyword displayed at the suggestedkeyword display part1403. When the customer attempts to perform commercial product search refinement based on the displayed suggested keyword, the customer inputs the suggested keyword or the suggested keyword and an additional keyword to the searchquery input part1401. Theinformation providing device100 performs searching again based on the input suggested keyword and the additional keyword, and displays the search result on the display of thecustomer terminal300. In addition, theinformation providing device100 may automatically perform search refinement when the customer selects any one of the suggested keywords displayed at the suggestedkeyword display part1403, not when the keyword is input to the searchquery input part1401 for search refinement.
• Note that the presentingunit123 presents, as the suggested keyword, the term with the calculated score that is at an upper ranking position, but the “upper ranking position term” is not limited to the one with the appearance frequency that is equal to or higher than the predetermined value like this embodiment. When, for example, among the relevant terms arranged in the order of the higher appearance frequency, the relevant terms with a ranking that is equal to or higher than a predetermined position may be taken as the upper ranking position terms. In addition, the upper ranking position term may be a term that has the appearance frequency as a predetermined relevant term at a total of equal to or greater than 60% as a whole.
• In this embodiment, thecontroller101 and thecommunicator105 work together to function as the presentingunit123.
• Next, an explanation will be given of an operation of theinformation providing device100 of this embodiment.
• An information providing process executed by thecontroller101 of theinformation providing device100 will be explained.FIG. 15 is a flowchart of an example information providing process. The information providing process illustrated inFIG. 15 starts upon receiving, for example, an inputting of a request to theranking page1400 of the commercial products from thecustomer terminal300. In addition, this process is executed by thecontroller101 that reads the program stored in theROM102.
• When receiving an inputting of a request to the commercialproduct ranking page1400 from thecustomer terminal300, the obtainingunit120 refers to the purchase table600, and obtains the ranking of the commercial products (step S101). The obtainingunit120 refers to, in accordance with to what category of commercial product the ranking request from thecustomer terminal300 corresponds, the purchase table600 in accordance with that category. In addition, the obtainingunit120 refers to the purchase table600, in accordance with to what ranking period, that is, real time, day by day, week by week, month by month, and year by year, the ranking request from thecustomer terminal300 corresponds.
• Next, the collectingunit121 collects the relevant terms to the commercial product for the upper ranking position commercial product obtained by the obtaining unit120 (step S102). The relevant term to the commercial product is collected from the terms contained in thesales page900 created by the shop or from the search query input by the customer.
• Subsequently, for each relevant term to the commercial product collected by the collectingunit121, the calculatingunit122 calculates the score of the term based on the number of collections of this term (step S103).
• Next, the presentingunit123 displays, on the display of thecustomer terminal300, the term with the upper ranking score among the scores of the terms calculated by the calculatingunit122 as the suggested keyword for commercial product search refinement(step S104). Subsequently, the process is finished.
• As explained above, theinformation providing device100 of this embodiment obtains the relevant terms to the commercial products at the upper positions in the commercial product ranking, and calculates the score based on the number of obtainments of such a term. Next, the term with the calculated score that is at an upper position can be presented as the suggested keyword for commercial product search refinement. Therefore, the customer can obtain the relevant term to the commercial product at the upper ranking position as the suggested keyword, and thus there is an advantage such that the customer can easily search a recommended commercial product at the upper position in the ranking by a search refinement based on the suggested keyword presented to the customer.
• The embodiment of the present disclosure was explained above, but the present disclosure is not limited to this embodiment. Modified examples of the embodiment will be explained below. In the following modified examples, the same structure as that of the above embodiment will be denoted by the same reference numeral, and the detailed explanation thereof will be omitted.
FIRST MODIFIED EXAMPLE OF FIRST EMBODIMENT
• In the above first embodiment, the calculatingunit122 calculates, based on the number of collections of the term, for a specific category, the score of the collected term. However, the calculatingunit122 can have a further limited calculation method of the score. For example, the score may be defined based on a value obtained by decreasing the number of collections of a specific term in a first category by the number of collections of such specific term in a second category.
• An explanation will be given below with reference to the relevant term appearance frequency table illustrated inFIG. 12 and for the category that is one-piece suit, and to the relevant term appearance frequency table illustrated inFIG. 13 and for the category that is dress. A specific term “free shipping” in a first category that is one-piece suit also appears in a second category that is dress. When the same term appears in multiple categories in this way, such a term is highly possibly a generic name or an industry term rather than a unique term to such a category. Hence, when there is such a term, it is necessary to decrease the score of such a term. In this modified example, the term “free shipping” appears in the category that is one-piece suit and in the category that is dress. Accordingly, the number of appearance frequency of the term “free shipping” in the category that is one-piece suit is decreased based on the appearance frequency of the term “free shipping” in the category that is dress.
• Possible methods to decrease the score are a method of simply subtracting the number of the appearance frequency of the term in the second category from the appearance frequency of the term in the first category, a method of multiplying a coefficient in accordance with the appearance frequency in the second category, and subtracting from the number of appearance frequency in the first category, and a method of subtracting a certain number from the number of appearance frequency in the first category when the term also appears in the second category.
• By employing such a structure, it is determined whether or not the collected relevant term to the commercial product is a unique term in the category to which such a term belongs, and thus the appropriate suggested keyword for commercial product search refinement can be extracted.
SECOND MODIFIED EXAMPLE OF FIRST EMBODIMENT
• In the above first modified example of the first embodiment, the calculatingunit122 defined the number of collections of the specific term in the first category based on the value decreased on the basis of the number of collections of the specific term in the second category. However, the degree of decreasing the value can be further limited. When, for example, the similarity between the first category and the second category is high, the decrease degree may be reduced.
• When the similarity between the categories is high, the term to be collected as the relevant term to the commercial product is common, and the number of collections of such a term increases. Hence, such a term is highly possibly a specific term to those categories. Conversely, when the similarity between the categories is low, the term to be collected as the relevant term to the commercial product is usually not so common. When the term to be collected between categories is common and the number of collections of such a term is high, such a term is highly possible a generic name or a common term. Therefore, it is preferable that the degree of decreasing the score of the term should be set to be low when the similarity between the categories is high.
• An explanation will be given with reference to the relevant term appearance frequency table illustrated inFIG. 12 and for the category that is one-piece suit, and to the relevant term appearance frequency table illustrated inFIG. 13 and for the category that is dress. A specific term “gift” in the first category that is one-piece suit also appears in the second category that is dress. Next, the category that is one-piece suit and the category that is dress are both the child categories of the parent category that is women's fashion, and thus both categories are highly possibly similar Hence, when there is such a term, and the score of such a term is to be decreased, it is necessary to reduce the amount of decrease. In addition, it is presumed that, for example, a term “long sleeve” in the category that is one-piece suit also appears in the category that is dress shirt in men's fashion. In this case, the parent category of the category that is one-piece suit is women's fashion, and the parent category of the category that is dress shirt is men's fashion. Hence, the similarity between both categories is lower than the similarity between the category that is one-piece suit and the category that is dress. Accordingly, when the score of the term is decreased, the amount of decrease is increased.
• By employing such a structure, when the score of the collected term is calculated, it is determined whether or not, in consideration of the similarity between a given category to which the collected relevant term to the commercial product belongs and the other category where such a term also appears, the term is a specific term to the given category, and thus the suggested keyword appropriate for commercial product search refinement can be extracted.
THIRD MODIFIED EXAMPLE OF FIRST EMBODIMENT
• In the above second modified example of the first embodiment, the calculatingunit122 defines the number of collections of the specific term in the first category based on the value decreased on the basis of the number of collections of the specific term in the second category, and when the similarity between the first category and the second category is high, the degree of decrease is reduced. However, as for the method of determining the similarity, the other method is also applicable. When, for example, the category hierarchy is defined based on a tree structure that has each category disposed at a node, and when a given category in the tree structure and the other category therein are brotherhood nodes, the similarity between the given category and the other category can be determined as high.
• More specifically, an explanation will be given with reference to thecategory tree structure401 illustrated inFIG. 4. As explained above, thetree structure401 inFIG. 4 is a tree structure for the category that is women's fashion, and has the node that is women's fashion as the root. This root has multiple child nodes that are one-piece suit, dress, and suit linked one another with edges. In thistree structure401, the relationship among the one-piece suit, dress, and suit satisfy the brotherhood relationship. In addition, the relationship among long, short, and medium in the category that is one-piece suit satisfy the brotherhood relationship. It can be determined that the similarity is high between those categories with the brotherhood relationship, and the decree of decreasing the score is reduced.
• By employing such a structure, when the score of the collected term is calculated, it is determined whether or not, in consideration of the node relationship between a given category to which the collected relevant term to the commercial product belongs and the other category where such a term also appears, the term is a specific term to the given category, and thus the suggested keyword appropriate for commercial product search refinement can be extracted.
FOURTH MODIFIED EXAMPLE OF FIRST EMBODIMENT
• In the above second modified example of the first embodiment, the calculatingunit122 defines the number of collections of the specific term in the first category based on the value decreased on the basis of the number of collections of the specific term in the second category, and when the similarity between the first category and the second category is high, the degree of decrease is reduced. However, as for the method of determining the similarity, the other method is also applicable. When, for example, the category hierarchy is defined based on a tree structure that has each category disposed at a node, and when a distance between a given category and the other category is short, it can be determined that the similarity between the given category and the other category is high.
• An explanation will be given below with reference to thecategory tree structure401 illustrated inFIG. 4. As explained above, thetree structure401 inFIG. 4 is a tree structure for the category that is women's fashion. In thetree structure401, from the child category that is long of the category that is one-piece suit to the child category that is medium of one-piece suit, it is necessary to pass through two edges that are long one-piece suit medium. Hence, the distance from long to medium can be calculated as the distance equivalent to the two edges. Conversely, from the child category that is long of the category that is one-piece suit to the child category that is party of dress, it is necessary to pass through four edges that are long→one-piece suit→women's fashion→dress→party. As explained above, the distance from the category which is long and is the child category of one-piece suit to the category that is medium is shorter than the distance from the category which is long and is the child category of one-piece suit to the child category that is party of dress. Hence, the similarity between the category that is long and the category that is medium can be determined as higher than the similarity between the category that is long and the category that is party. Next, the degree of decreasing the score is reduced.
• By employing such a structure, when the score of the collected term is calculated, it is determined whether or not, in consideration of the distance from a given category to which the collected relevant term to the commercial product belongs to the other category where such a term also appears, the term is a specific term to the given category, and thus the suggested keyword appropriate for commercial product search refinement can be extracted.
FIFTH MODIFIED EXAMPLE OF FIRST EMBODIMENT
• In the above first embodiment, the obtainingunit120 obtains the ranking of the commercial products belonging to a specific category in the e-marketplace. However, the obtainment of the commercial product ranking can be further limited in consideration of the relationship between the commercial product and the shop that sells the commercial product. For example, in the sales pages created by the multiple shops, when the same commercial product belonging to the same category is sold, regardless of the consistency/inconsistency of the shop, the ranking may be obtained for each category where the sold commercial product belongs.
• In the e-marketplace, since commercial product names are created by the shops, the same commercial product belonging to the same category is often available from the multiple shops with different commercial product names. In this case, theinformation providing device100 creates the purchase table600 with those commercial products being as different commercial products, and the obtainingunit120 obtains the ranking from this purchase table600 created in this way. Hence, many commercial products are on sale although such commercial products are the same commercial product in practice, and such commercial products are taken as the commercial product at the lower position in the ranking even if the large number of such a commercial product has been sold.
• In order to avoid this inconvenience, this modified example is applied. More specifically, in thecommercial product database400, for the same commercial product available in the same category, for example, a commercial product code is additionally given to handle those commercial products as the same commercial product. When, for example, in thecommercial product database400, the multiple commercial products that are “A-line one-piece suit” and the “frilly one-piece suit” belonging to the same category (J011) are given with the same commercial product code, those commercial products are taken as the same commercial product. Hence, the commercial product with the commercial product ID “M0001” in the purchase table600 and the commercial product with the commercial product ID “M0002” therein are taken as the same commercial product, and when the obtainingunit120 creates the ranking table800, the total amount of sales is combined and calculated. Hence, the ranking in the ranking table inFIG. 8, in general, the first position is the commercial product M004, the second position is the commercial product M002, and the third position is the commercial product M001. According to this modified example, however, the first position becomes a commercial product group including M001 and M002, and the second position is the commercial product M004.
• By employing such a structure, even if the multiple shops are separately selling the same commercial product, it is possible to present the suggested keyword of the relevant term to the commercial product with reference to the ranking based on the same commercial product. Hence, the customer can view, as the suggested keyword, the search term with a larger appearance frequency regardless of the shops that are the sources of the commercial product, and can perform effective search refinement.
Second Embodiment
• Next, an explanation will be given of a second embodiment. In the first embodiment, the score of the relevant term to the commercial product is calculated based on the number of collections of the term, and the suggested keyword for commercial product search refinement is determined. In this embodiment, as for the calculation of the score, the number of collections of the term is multiplied by a contrary category frequency, and the score is determined.
• The contrary category frequency is a reference index indicating the rate of the appearance frequency of the term in accordance with the number of categories, and the larger the number of categories where the term appears is, the smaller the contrary category frequency becomes. The contrary category frequency can be expressed as the following formula.
• 
  Contrary category frequency=log(number of child categories/number of appearing child categories)
• The calculation of the score using the contrary category frequency is applicable to the category hierarchy that is a tree structure with multiple categories disposed at nodes. In general, the category hierarchy expressed by the tree structure is the category hierarchy with thetree structure401 illustrated inFIG. 4, but in this embodiment, in order to facilitate understanding to the present disclosure, an explanation will be given based on a simpler category hierarchy illustrated inFIG. 16.
• More specifically, in the category hierarchy of the tree structure with the multiple categories disposed at nodes, the calculatingunit122 calculates the appearance frequency of the relevant term in the child category and in the parent category, and the contrary category frequency is calculated based on the calculated appearance frequency.
• For example, in the parent category that is women's fashion inFIG. 16, and in the category hierarchy where nodes of child categories that are one-piece suit and dress are disposed, the appearance frequency of the relevant term for one-piece suit is illustrated inFIG. 12 and the appearance frequency of the relevant term for dress is illustrated inFIG. 13. Based on the appearance frequency of the relevant term illustrated in such figures, the calculatingunit122 obtains the appearance frequency of the relevant term in women's fashion.FIG. 17 is a table showing an example appearance frequency of the relevant term in women's fashion.
• The calculatingunit122 obtains the contrary category frequency based on the appearance frequency of the relevant term illustrated inFIG. 17.FIG. 18 illustrates an example table showing a relationship between the relevant term and the contrary category frequency. The table illustrated inFIG. 18 and showing the relationship between the relevant term and the contrary category frequency includes the fields that are the relevant term, the appearance frequency of the relevant term, the number of child categories, the number of categories where the term appears, and the contrary category frequency. The values of the appearance frequency, number of child categories, and number of categories where the term appears are substituted in the above formula to obtain the contrary category frequency. For example, as illustrated inFIG. 18, the calculatingunit122 calculates, by applying the above formula, the contrary category frequency of the relevant term as “0” for “free shipping” and “gift”, and as “0.301” for other relevant terms. When the number of categories where the term appears is zero, the contrary category frequency is set to be, for example, 1, and the calculation is performed.
• Next, the calculatingunit122 calculates the weighting of each relevant term based on the product of the obtained contrary category frequency by the appearance frequency.FIG. 19 illustrates an example table for weighting calculated for each relevant term. The weighting table includes the relevant term, the appearance frequency, the contrary category frequency, and the weighting value obtained by the appearance frequency×contrary category frequency. The presentingunit123 presents, to thecustomer terminal300, the term with the upper position weighting value among the weighting values calculated in this way as the suggested keyword for commercial product search refinement.
• When there are multiple terms which have the same meaning but have different notation, those terms can be processed as the same relevant term. More specifically, when a similarity between a word X and a word Y is smaller than a threshold k, the calculatingunit122 processes that the word X and the word Y has the same meaning. As for the multiple terms recognized as the same term, a representative term is taken as the same term. For example, the term “funeral ceremony” and the term “funeral” are taken as the same relevant terms as the similarity is smaller than the threshold k, and the term “funeral” is taken as the representative term.
• As explained above, for the term relevant to the commercial product and presented as the keyword for commercial product search refinement, after the presentingunit123 obtains the terms with the number of appearance frequency at the higher ranking position, the ranking position is changed based on the contrary category frequency, and similar terms can be further collected as a single representative term. Through such procedures, a highly accurate suggested keyword for commercial product search refinement can be presented.
• FIGS. 20A to 20C are diagrams illustrating how the suggested keywords for commercial product search refinement are narrowed down.FIG. 20A is a diagram illustrating a table that arranges the relevant terms to the commercial products in the order of the appearance frequency in the category that is women's fashion,FIG. 20B is a diagram illustrating a table that changes the arrangement of the relevant terms to the commercial products with weighting based on the contrary category frequency, and FIG.
• 20C is a diagram illustrating a table showing the similar terms collected as a term for the relevant terms to the commercial products. As illustrated inFIG. 20A andFIG. 20B, the relevant term that is “free shipping” is arranged with the appearance frequency being as the first ranking position, but after the weighting based on the contrary category frequency, this term is located at the lower ranking position. In addition, “funeral ceremony” and “funeral” are recognized as the same word, and are collectively displayed by the representative term that is “funeral”. Eventually, as illustrated inFIG. 20C, the terms not specific to this category are once arranged to the lower ranking positions, and the multiple terms with the same meaning are collectively displayed as a single term.
• By employing such a structure, the multiple relevant terms to the commercial product in a given category are narrowed down to the specific term to such a category based on the contrary category frequency, and the terms with the similar meaning are collected up by a representative term. Hence, a further precise suggested keyword can be presented to the customer, and the customer can apply such a suggested keyword for search refinement.
• In the above embodiments, the program to be executed by theinformation providing device100 may be stored in a non-transitory computer-readable recording medium, such as a flexible disk, a Compact Disk Read-Only Memory (CD-ROM), a Digital Versatile Disk (DVD), or an Magneto to Optical Disk (MO) and distributed. Next, by installing such a program to an information processing device like a personal computer to realize theinformation providing device100 that executes the above process.
• In addition, when the above functions are realized and shared by an
• Operating System (OS) or by a cooperative work of the OS with an application, only a program other than a portion that realizes the functions of the OS may be stored in a non-transitory recording medium and distributed or may be downloaded.
• Preferred embodiments of the present disclosure were explained above, but the present disclosure is not limited to the specific embodiment, and various changes and modifications can be made within the scope of the appended claims.
• In addition, the above embodiments are to explain the present disclosure, and are not intended to limit the scope of the present disclosure. That is, the scope of the present disclosure is indicated by the appended claims rather than the embodiments. In addition, various modifications carried out within the appended claims and the equivalent range thereto should be determined as being within the scope of the present disclosure.
INDUSTRIAL APPLICABILITY
• The present disclosure is suitable for e-commerce that utilizes a network like the Internet.
REFERENCE SIGNS LIST
• 1 Information providing system
• 100 Information providing device
• 101 Controller
• 102 ROM
• 103 RAM
• 104 Display
• 105 Communicator
• 106 Operation hardware
• 107 Bus
• 120 Obtaining unit
• 121 Collecting unit
• 122 Calculating unit
• 123 Presenting unit
• 200 Shop terminal
• 300 Customer terminal
• 400 Commercial product database
• 401 Tree structure
• 500 Internet
• 600 Purchase table
• 700 Category table
• 800 Ranking table
• 900-902 Sales page
• 900aCommercial product description display part
• 1000 Relevant term table
• 1100 Relevant term table
• 1200 Relevant term appearance frequency table
• 1300 Relevant term appearance frequency table
• 1400 Ranking table
• 1401 Search query input part
• 1402 Ranking time period display part
• 1403 Suggested keyword display part
• 1404 Ranking display part
• 1700 Relevant term appearance frequency table

Claims (12)

1. An information providing device comprising:

an obtaining unit that obtains a ranking of a commercial product belonging to a category defined on an e-marketplace;

a collecting unit that collects a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;

a calculating unit that calculates a score of each collected term based on at least a number of collections of the term; and

a presenting unit that presents, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position, wherein:

in the e-marketplace, a category hierarchy is defined by a tree structure with each category disposed at a node; and

the score of the collected term for the category is defined based on a product of:

(a) a weighting defined by the number of collections of the term; and

(b) a contrary category frequency defined by a number of child categories of the category, and a number of child categories where the term is collected among the child categories.

2. (canceled)

3. An information providing device comprising:

an obtaining unit that obtains of a commercial product belonging to a category defined on a e-marketplace;

a collecting unit that collects a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;

a calculating unit that calculates a score of each collected term based 5 on at least a number of collections of the term; and

a presenting unit that presents, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position, wherein:

the score of the collected term for the category is defined based on a value obtained by decreasing the number of collections of the term for the category on a basis of a number of collections of the term in another category.

4. The information providing device according toclaim 3, wherein:

a higher a similarity between the category and the other category is, a smaller a degree of decrease is.

5. The information providing device according toclaim 3, wherein:

in the e-marketplace, a category hierarchy is defined by a tree structure with each category disposed at a node; and

in the tree structure, when the category and the other category are brotherhood categories, a similarity between the category and the other category is high.

6. The information providing device according toclaim 3, wherein:

in the e-marketplace, a category hierarchy is defined by a tree structure with each category disposed at a node; and

in the tree structure, when a distance between the category to the other category is short, a similarity between the category and the other category is high.

7. The information providing device according toclaim 1, wherein:

in the e-marketplace, each of a plurality of shops creates the sales page, and the same commercial product belonging to the same category are available from the respective sales pages; and

the ranking is obtained for each category to which the sold commercial product belongs regardless of from which shop the commercial product is sold.

8. An information providing method executed by an information providing device including an obtaining unit, a collecting unit, a calculating unit, and a presenting unit, the method comprising:

obtaining by the obtaining unit a ranking of a commercial product belonging to a category defined on an e-marketplace;

collecting by the collecting unit a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;

calculating by the calculating unit a score of each collected term based on at least a number of collections of the term; and

presenting by the presenting unit, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position, wherein:

in the e-marketplace, a category hierarchy is defined by a tree structure with each category disposed at a node; and

the score of the collected term for the category is defined based on a product of :

(a) a weighting defined by the number of collections of the term; and

(b) a contrary category frequency defined by a number of child categories of the category, and a number of child categories where the term is collected among the child categories.

9. (canceled)

10. (canceled)

11. An information providing method executed by an information providing device including an obtaining unit, a collecting unit, a calculating unit, and a presenting unit, the method comprising:

obtaining by the obtaining unit a ranking of a commercial product belonging to a category defined on an e-marketplace;

collecting by the collecting unit a term relevant to a commercial product at an upper position in the obtained ranking from a text contained in a sales page for selling the upper ranking position commercial product or a search query that triggers the sales page to be viewed;

calculating by the calculating unit a score of each collected term based on at least a number of collections of the term; and

presenting by the presenting unit, as a suggested keyword for search refinement of the commercial product belonging to the category, the term with the calculated score at an upper ranking position, wherein:

the score of the collected term for the category is defined based on a value obtained by decreasing the number of collections of the term for the category on a basis of a number of collections of the term in another category.

12. The information providing device according toclaim 1, wherein:

the score of the collected term for the category is defined based on a value obtained by decreasing the number of collections of the term for the category on a basis of a number of collections of the term in another category.

Images